by Clay Shirky
Post-Managerial Organization
When an organization takes on a task, the difficulty of coordinating everyone needs to be reined in somehow, and the larger the group, the more urgent the need. The standard, almost universal solution is to create a hierarchy and to slot individuals into that organization by role. In Coasean terms, McCallum’s system lowered the transaction costs of running a railroad by increasing managerial structure. This approach greatly simplifies lines of responsibility and communication, making even very large organizations manageable. The individuals in such an organization have to agree to be managed, of course, which is usually achieved by paying them, and by making continued receipt of their pay contingent on their responsiveness to their manager’s requests.
An organization will tend to grow only when the advantages that can be gotten from directing the work of additional employees are less than the transaction costs of managing them. Coase concentrated his analysis on businesses, but the problems of coordination costs apply to institutions of all sorts. The Catholic Church and the U.S. Army are as hierarchical as any for-profit business, and for many of the same reasons. The layers of structure between the pope and the priests, or between the president and the privates, is a product of the same forces as the layers between the general superintendent and a conductor on the New York & Erie. This hierarchical organization reduces transaction costs, but it doesn’t eliminate them.
Imagine a company with fifteen hundred employees, where each manager is responsible for half a dozen people. The CEO has six vice presidents, who each direct the work of six supervisors, and so on. Such a company would have three layers of management between the boss and the workers. If you want to bring the workers closer to the boss, you will have to increase the number of workers that each manager is responsible for. This will reduce the number of layers but will also reduce average management time with each staff member (or force everyone to spend more hours per day communicating with one another). When an organization grows very large, it reaches the limit implicit in Coase’s theory; at some point an institution simply cannot grow anymore and still remain functional, because the cost of managing the business will destroy any profit margin. You can think of this as a Coasean ceiling, the point above which standard institutional forms don’t work well.
Coase’s theory also tells us about the effects of small changes in transaction costs. When such costs fall moderately, we can expect to see two things. First, the largest firms increase in size. (Put another way, the upper limit of organizational size is inversely related to management costs.) Second, small companies become more effective, doing more business at lower cost than the same company does in a world of high transaction costs. These two effects describe the postwar industrial world well: Giant conglomerates like ITT in the 1970s and GE in recent years used their management acumen to get into a huge variety of businesses, simply because they were good at managing transaction costs. At the same time there has been an explosion of small- and medium-sized businesses, because such businesses were better able to discover and exploit new opportunities.
But what if transactions costs don’t fall moderately? What if they collapse? This scenario is harder to predict from Coase’s original work, and it used to be purely academic. Now it’s not, because it’s happening, or rather it has already happened, and we’re starting to see the results.
Anyone who has worked in an organization with more than a dozen employees recognizes institutional costs. Anytime you are faced with too many meetings, too much paperwork, or too many layers of approval (shades of McCallum), you are dealing with those costs. Until recently, such costs have been little more than the stuff of water-cooler grumbling—everyone complains about institutional overhead, without much hope of changing things. In that world (the world we lived in until recently), if you wanted to take on a task of any significance, managerial oversight was just one of the costs of doing business.
What happens to tasks that aren’t worth the cost of managerial oversight? Until recently, the answer was “Those things don’t happen.” Because of transaction costs a long list of possible goods and services never became actual goods and services; things like aggregating amateur documentation of the London transit bombings were simply outside the realm of possibility. That collection now exists because people have always desired to share, and the obstacles that prevented sharing on a global scale are now gone. Think of these activities as lying under a Coasean floor; they are valuable to someone but too expensive to be taken on in any institutional way, because the basic and unsheddable costs of being an institution in the first place make those activities not worth pursuing.
Our basic human desires and talents for group effort are stymied by the complexities of group action at every turn. Coordination, organization, even communication in groups is hard and gets harder as the group grows. That difficulty means that whatever methods help coordinate group action will spread, no matter how inefficient they are, so long as they are better than nothing. Small groups have several methods for coordinating action, like calling each group member in turn or setting up a phone tree, but most of these methods don’t work well even for dozens of people, much less for thousands. For large-scale activity, the methods that have worked best have been those pioneered by McCallum—hierarchical organization, managed in layers. The most common organizational structures we have today are simply the least bad fit for group action in an environment of high transaction costs.
Our new tools offer us ways of organizing group effort without resorting to McCallum’s strategies. Flickr stands in a different kind of relationship to its photographers than a newspaper does. Where a newspaper is in the business of directing the work of photographers, Flickr is simply a platform; whatever coordination happens comes from the users and is projected onto the site. This is odd. We generally regard institutions as being capable of more things than uncoordinated groups are, precisely because they are able to direct their employees. Here, though, we have a situation where the loosely affiliated group can accomplish something more effectively than the institution can. Thanks to the introduction of user-generated labeling, the individual motivation of the photographers—devoid of financial reward—is now enough to bring vast collections of photos into being. These collections didn’t just happen to be put together without an institution; that is the only way they could have been put together.
This is where Coasean logic gets strange. Small decreases in transaction costs make businesses more efficient, because the constraints of the institutional dilemma get less severe. Large decreases in transaction costs create activities that can’t be taken on by businesses, or indeed by any institution, because no matter how cheap it becomes to perform a particular activity, there isn’t enough payoff to support the cost incurred by being an institution in the first place. So long as the absolute cost of organizing a group is high, unmanaged groups will be limited to undertaking small efforts—a night out at the movies, a camping trip. Even something as simple as a potluck dinner typically requires some hosting institution. Now that it is possible to achieve large-scale coordination at low cost, a third category has emerged: serious, complex work, taken on without institutional direction. Loosely coordinated groups can now achieve things that were previously out of reach for any other organizational structure, because they lay under the Coasean floor.
The cost of all kinds of group activity—sharing, cooperation, and collective action—have fallen so far so fast that activities previously hidden beneath that floor are now coming to light. We didn’t notice how many things were under that floor because, prior to the current era, the alternative to institutional action was usually no action. Social tools provide a third alternative: action by loosely structured groups, operating without managerial direction and outside the profit motive.
From Sharing to Cooperation to Collective Action
For the last hundred years the big organizational question has been whether any given task was best taken on by the state, di
recting the effort in a planned way, or by businesses competing in a market. This debate was based on the universal and unspoken supposition that people couldn’t simply self-assemble; the choice between markets and managed effort assumed that there was no third alternative. Now there is. Our electronic networks are enabling novel forms of collective action, enabling the creation of collaborative groups that are larger and more distributed than at any other time in history. The scope of work that can be done by noninstitutional groups is a profound challenge to the status quo.
The collapse of transaction costs makes it easier for people to get together—so much easier, in fact, that it is changing the world. The lowering of these costs is the driving force underneath the current revolution and the common element to everything in this book. We’re not used to thinking of “groupness” as a specific category—the differences between a college seminar and a labor union seem more salient than their similarities. It’s hard to see how Evan Guttman’s quest for the return of the mobile phone is the same kind of thing as the distributed documentation of the Indian Ocean tsunami. But like a chain of volcanoes all fed by the same pool of magma, the surface manifestations of group efforts seem quite separate, but the driving force of those eruptions is the same: the new ease of assembly. This change can be looked at as one long transition, albeit one with many manifestations, unfolding at different speeds in different contexts. The transition can be described in basic outline as the answer to two questions: Why has group action largely been limited to formal organizations? What is happening now to change that?
We now have communications tools—and increasingly, social patterns that make use of those tools—that are a better fit for our native desires and talents for group effort. Because we can now reach beneath the Coasean floor, we can have groups that operate with a birthday party’s informality and a multinational’s scope. What we are seeing, in the amateur coverage of the Thai coup and the tsunami documentation and the struggle over Ivanna’s phone and countless other examples, is the beginning of a period of intense experimentation with these tools. The various results look quite different from one another, and as we get good at using the new tools, those results will diverge still further. New ease of assembly is causing a proliferation of effects, rather than a convergence, and these effects differ by how tightly the individuals are bound to one another in the various groups.
You can think of group undertaking as a kind of ladder of activities, activities that are enabled or improved by social tools. The rungs on the ladder, in order of difficulty, are sharing, cooperation, and collective action.
Sharing creates the fewest demands on the participants. Many sharing platforms, such as Flickr, operate in a largely take-it-or-leave-it fashion, which allows for the maximum freedom of the individual to participate while creating the fewest complications of group life. Though Flickr sets public sharing as the default, it also allows users to opt to show photos only to selected users, or to no one. Knowingly sharing your work with others is the simplest way to take advantage of the new social tools. (There are also ways of unknowingly sharing your work, as when Google reads the linking preferences of hundreds of millions of internet users. These users are helping create a communally available resource, as Flickr users are, but unlike Flickr, the people whose work Google is aggregating aren’t actively choosing to make their contributions.)
Cooperation is the next rung on the ladder. Cooperating is harder than simply sharing, because it involves changing your behavior to synchronize with people who are changing their behavior to synchronize with you. Unlike sharing, where the group is mainly an aggregate of participants, cooperating creates group identity—you know who you are cooperating with. One simple form of cooperation, almost universal with social tools, is conversation; when people are in one another’s company, even virtually, they like to talk. Sometimes the conversation is with words, as with e-mail, IM, or text messaging, and sometimes it is with other media: YouTube, the video sharing site, allows users to post new videos in response to videos they’ve seen on the site. Conversation creates more of a sense of community than sharing does, but it also introduces new problems. It is famously difficult to keep online conversations from devolving into either name-calling or blather, much less to keep them on topic. Some groups are perfectly happy with those effects (indeed, there are communities on the internet that revel in puerile or fatuous conversation), but for any group determined to maintain a set of communal standards some mechanism of enforcement must exist.
Collaborative production is a more involved form of cooperation, as it increases the tension between individual and group goals. The litmus test for collaborative production is simple: no one person can take credit for what gets created, and the project could not come into being without the participation of many. Structurally, the biggest difference between information sharing and collaborative production is that in collaborative production at least some collective decisions have to be made. The back-and-forth talking and editing that makes Wikipedia work results in a single page on a particular subject (albeit one that changes over time). Collaboration is not an absolute good—many tools work by reducing the amount of required coordination, as Flickr does in aggregating photos. Collaborative production can be valuable, but it is harder to get right than sharing, because anything that has to be negotiated about, like a Wikipedia article, takes more energy than things that can just be accreted, like a group of photos.
Collective action, the third rung, is the hardest kind of group effort, as it requires a group of people to commit themselves to undertaking a particular effort together, and to do so in a way that makes the decision of the group binding on the individual members. All group structures create dilemmas, but these dilemmas are hardest when it comes to collective action, because the cohesion of the group becomes critical to its success. Information sharing produces shared awareness among the participants, and collaborative production relies on shared creation, but collective action creates shared responsibility, by tying the user’s identity to the identity of the group. In historical terms, a potluck dinner or a barn raising is collaborative production (the members work together to create something), while a union or a government engages in collective action, action that is undertaken in the name of the members meant to change something out in the world, often in opposition to other groups committed to different outcomes.
The commonest collective action problem is described as the “Tragedy of the Commons,” biologist Garrett Hardin’s phrase for situations wherein individuals have an incentive to damage the collective good. The Tragedy of the Commons is a simple pattern to explain, and once you understand it, you come to see it everywhere. The standard illustration of the problem uses sheep. Imagine you are one of a group of shepherds who graze their sheep on a commonly owned pasture. It’s obviously in everyone’s interest to keep the pasture healthy, which would require each of you to take care that your sheep don’t overgraze. As long as everyone refuses to behave greedily, everyone benefits. There is just one problem with this system: “everyone” doesn’t take your sheep to market. You do. Your incentive, as an individual shepherd, is to minimize the cost of raising the fat-test possible sheep. Everyone benefits from you moderating your sheep’s consumption of grass, but you would benefit more from free riding, which is to say letting them eat as much free grass as they possibly could.
Once you have this realization, you can still refrain from what would ultimately be a ruinous strategy, on the grounds that it would be bad for everyone else. Then another, even more awful thought strikes you: every other shepherd will have the same realization, and if even one of them decides to overgraze, all your good works will only end up subsidizing them. Seen in this light, the decision not to overgraze is provisional on everyone else making the same decision, which makes it very fragile indeed. The minute one of the other shepherds keeps his sheep out in the pasture an hour longer than necessary, the only power you have is to retaliate by doing the same. And this is
the Tragedy of the Commons: while each person can agree that all would benefit from common restraint, the incentives of the individuals are arrayed against that outcome.
People who benefit from a resource while doing nothing in recompense are free riders. Societies have generally dealt with the problem of free riders in one of two ways. The first way is elimination of the commons, transferring ownership of parts of it to individuals, all of whom have an incentive to protect their own resources. If six shepherds each own one-sixth of the former commons, the overgrazing problem is a personal one, not a social one. If you overgraze your section, you will suffer the future consequences, while your neighbor will not. The second way is governance or, as Hardin puts it, “mutual coercion, mutually agreed upon.” This solution prevents the individual actors from acting in their own interests rather than in the interests of the group. The Tragedy of the Commons is why taxes are never voluntary—people would opt out of paying for road maintenance if they thought their neighbors would pay for it. It’s also why restaurants often add an automatic tip for large parties—when enough people are eating, everyone feels comfortable underfunding the group’s tip, even if only unconsciously.
Collective action involves challenges of governance or, put another way, rules for losing. In any group that is determined to take collective action, different members of the group will express different opinions. Whenever a decision is taken on behalf of the group, at least some members won’t get their way, and the bigger the group is, or the more decisions are made, the more often this will happen. For a group to take collective action, it must have some shared vision strong enough to bind the group together, despite periodic decisions that will inevitably displease at least some members. For this reason collective action is harder to arrange than information sharing or collaborative creation. In the current spread of social tools, real examples of collective action—where a group acts on behalf of, and with shared consequences for, all of its members—are still relatively rare.
